Chapter 4. Planning a Hadoop Cluster
Picking a Distribution and Version of Hadoop
One of the first tasks to take on when planning an Hadoop deployment is selecting the distribution and version of Hadoop that is most appropriate given the features and stability required. This process requires input from those that will eventually use the cluster: developers, analysts, and possibly other systems such as business intelligence applications. This isn’t dissimilar from selecting a relational database based on what is required by downstream applications. For instance, some relational databases support extensions to SQL for advanced analytics, while other support features such as table partitioning in order to scale large tables or improve query performance.
Hadoop is, as previously mentioned, an Apache Software Foundation (ASF) project. This means it’s available directly from Apache in both source and binary formats. It’s extremely common though for people to use more than just core Hadoop. While Hadoop is absolutely critical—after all, it provides not just the distributed file-system, but also the MapReduce processing framework—many users view it as the core of a larger system. In this sense, Hadoop is analogous to an operating system kernel, giving us the core functionality upon which we build higher-level systems and tools. Many of these related libraries, tools, languages, and systems are also open source projects available from the ASF.
There is an inherent complexity in assembling these ...
Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.
Read now
Unlock full access